U.S. patent application number 10/278515 was filed with the patent office on 2003-05-01 for decoder and receiver using the same.
Invention is credited to Hattori, Toshihiro, Morita, Hideyuki, Uegaki, Tsukie.
Application Number | 20030084392 10/278515 |
Document ID | / |
Family ID | 19143135 |
Filed Date | 2003-05-01 |
United States Patent
Application |
20030084392 |
Kind Code |
A1 |
Hattori, Toshihiro ; et
al. |
May 1, 2003 |
Decoder and receiver using the same
Abstract
A turbo decoder has a control portion, a decoding portion and
the like. The control portion determines the repeating number of
times of the calculation for the second likelihood Ra to obtain the
repeating number of times in association with the condition of the
communication environment. The decoding portion decodes the
error-correction code by using the repeating number of times
determined at the control portion so that the decoded data is
obtained efficiently by performing the decode so as to correspond
to the condition of the communication environment.
Inventors: |
Hattori, Toshihiro;
(Okazaki-City, JP) ; Morita, Hideyuki;
(Nishio-City, JP) ; Uegaki, Tsukie; (Kariya-City,
JP) |
Correspondence
Address: |
HARNESS, DICKEY & PIERCE, P.L.C.
P.O. BOX 828
BLOOMFIELD HILLS
MI
48303
US
|
Family ID: |
19143135 |
Appl. No.: |
10/278515 |
Filed: |
October 23, 2002 |
Current U.S.
Class: |
714/752 |
Current CPC
Class: |
H03M 13/6337 20130101;
H03M 13/6577 20130101; H04L 1/005 20130101; H04L 1/0051 20130101;
H03M 13/2957 20130101; H04L 1/0066 20130101; H04L 1/20 20130101;
H04L 1/0036 20130101; H03M 13/3707 20130101 |
Class at
Publication: |
714/752 |
International
Class: |
H03M 007/20; H03M
013/00 |
Foreign Application Data
Date |
Code |
Application Number |
Oct 24, 2001 |
JP |
2001-326818 |
Claims
What is claimed is:
1. A decoder for being adopted to a receiver receiving a signal
having an error-correction code and a known code of an information
code comprising: a decoding portion for decoding the received
error-correction code using a parameter; an error calculating
portion for calculating an error level of the known code; and a
determining portion for determining said parameter based on the
error level calculated at the error calculating portion.
2. A decoder according to claim 1, further comprising: a
determining means for determining whether or not the calculated
error level is equal to or higher than a reference value, wherein
said determining portion includes: a first means for selecting a
first value as said parameter when it is determined that the error
level is equal to or higher than the reference value; a second
means for selecting a second value so that the precision in
decoding is lower than that in decoding by using the first value as
the parameter and a process speed in decoding is higher than that
in decoding by using the first value as the parameter when it is
determined that the error level is lower than the reference
value.
3. A decoder according to claim 2, wherein: the decoding portion
repeats a calculation for obtaining the likelihood of said
information code for a predetermined times by using said
error-correction code and the previous likelihood of the
information code obtained at a previous time, and obtains the
decoded data based on the likelihood of the information code
obtained by the calculation repeated for the predetermined times,
and: said determining portion determines said predetermined times
as the parameter based on the calculated error level.
4. A decoder according to claim 3, wherein: said first means
selects a first number of times as said predetermined times when it
is determined that the error level is equal to or higher than the
reference value, and: said second means selects a second number of
times smaller than the first number of times as said predetermined
times when it is determined that the error level is lower than the
reference value.
5. A decoder for being adopted to a receiver receiving a signal
having an error-correction code and a known code of an information
code, and converting the error-correction code to an N-bit digital
signal, comprising: an error calculating portion for calculating an
error level of the known code; a determining means for determining
whether or not the calculated error level is equal to or higher
than a reference value; a data setting means for obtaining the data
having a constant value in the L-bit portion of the N-bit digital
signal data from a LSB side of the N-bit digital signal, wherein a
value of the L in the L-bit is smaller than that of the N in the
N-bit; and a decoding portion for decoding the N-bit digital signal
when it is determined that the error level is equal to or higher
than the reference value, and decoding the data obtained at the
data setting means when it is determined that the error level is
lower than the reference value.
6. A decoder for being adopted to a receiver receiving a signal
having an error-correction code and a known code of an information
code, and converting the error-correction code to an N-bit digital
signal, comprising: an error calculating portion for calculating an
error level of the known code; a determining means for determining
whether or not the calculated error level is equal to or higher
than a reference value; a data modifying means for modifying the
N-bit digital data to a M-bit digital data, wherein the value of
the M in the M-bit is smaller than the value of the N in the N-bit;
and a decoding portion for decoding the N-bit digital signal when
it is determined that the error level is equal to or higher than
the reference value, and decoding the M-bit digital signal when it
is determined that the error level is lower than the reference
value.
Description
CROSS REFERENCE TO RELATED APPLICATION
[0001] This application is based upon Japanese Patent Application
No. 2001-32618, filed on Oct. 24, 2001, the contents of which are
incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to a decoder for decoding an
error-correcting code and a receiver using the same.
[0004] 2. Related Art
[0005] A code error occurs easily in the mobile communication
system. Therefore, a turbo encoder and a turbo decoder are
suggested to be used.
[0006] The turbo encoder is employed in a radio wave transmitter to
produce an error-correcting code of a information symbol based on
the block data having plural information symbols as the information
code. The turbo decoder obtains likelihood in each block data based
on the error-correction code sent from the radio wave transmitter,
and obtains the decode data based on the likelihood.
[0007] As shown in FIG. 1, the turbo encoder has a recursive
convolution encoders 10 and 20, an interleaver 30, and a
parallel/serial converter 40.
[0008] The block data constituted by the information symbols J(i)
to J(i+2) ( "i" denotes an order of the information symbols) is
input to the turbo encoder. The recursive convolution encoder 10
obtains a recursive organized code S(i) based on each information
symbol in the block data.
[0009] The recursive organized code S(i) is constituted by the
information symbol J(i), the delayed information symbol that is
delayed for a predetermined time with respect to the information
symbol J(i), and the further delayed information symbol that is
delayed for a predetermined time with respect to the delayed
information symbol.
[0010] The interleaver 30 changes the order of the information
symbols J(i) to J(i+2) and outputs them. The recursive convolution
encoder 20 obtains a recursive organized code U(i) based on the
information symbols output from the interleaver 30.
[0011] The recursive organized code U(i) is constituted by the
information symbol U(i), the delayed information symbol that is
delayed for a predetermined time with respect to the information
symbol U(i), and the further delayed information symbol that is
delayed for a predetermined time with respect to the delayed
information symbol.
[0012] The parallel/serial converter 40 is input with the
information symbols constituting the block data as a organized code
J(i), the recursive organized code S(i) and recursive organized
code U(i). The parallel/serial converter 40 converts the organized
code J(i), the recursive organized code S(i) and recursive
organized code U(i) into error-correction codes of the serial data.
The error-correction code is transmitted to the radio wave
transmitter by being produced one after another from all of the
information symbols in the block data.
[0013] Next, a turbo decoder will be described with reference to
FIGS. 2 and 3.
[0014] The turbo decoder has a control portion 50 and a decoding
portion 60. The decoding portion 60 obtains the likelihood of the
information symbol, i.e., the decoded data. The decoding portion 60
has a serial/parallel converter 61, soft-determination decoders 62
and 63, an interleavers 64 and 65, a deinterleaver 66 and a soft
determining portion 67. The likelihood means the probability of "1"
being sent as the information symbol, or the probability of "0"
being sent as the information symbol.
[0015] The serial/parallel converter 61, the soft-determination
decoders 62 and 63, the interleavers 64 and 65, the deinterleaver
66 and the soft determining portion 67 are activated in accordance
with direction sent from the control portion 50, respectively.
These activations will be described with reference to FIG. 3.
[0016] First of all, the serial/parallel converter 61 performs a
serial/parallel conversion of the error-correction code transmitted
from the radio wave transmitter so as to output the organized code
Ja(i) (=Ja(1), Ja(2), Ja(3)), the recursive organized code Sa(i)
(=Sa(1), Sa(2), Sa(3)) and the recursive organized code Ua(i)
(=Ua(1), Ua(2), Ua(3)) successively.
[0017] Next, the interleaver 64 changes the order of the organized
code Ja(1)-Ja(3) so as to correspond to the order of the recursive
organized code Ua(1)-Ua(3) at step 100. Hereinafter, the
order-changed organized code Ja(1)-Ja(3) are referred to as
organized code Jb(1), Jb(2), Jb(3), respectively.
[0018] After that, the calculation for the first likelihood P for
the block data by using the soft-determination decoder 62 at steps
110 and 120, the order-change of the first likelihood P by using
the interleaver 64 at step 130, the calculation for the second
likelihood R for the block data by using the soft-determination
decoder 63 at steps 140 and 150 and the order-change of the second
likelihood R by using the interleaver 64 at step 160 are performed
repeatedly for (X-1) times to obtain the second likelihood Ra(1,
X-1), Ra(2, X-1), Ra(3, X-1). The flow will be followed by the X
times calculation.
[0019] The soft-determination decoder 62 obtains the first
likelihood P(1, X) based on the organized code Ja(1), the recursive
organized code Sa(1) and the second likelihood Ra(1, X-1) at the
(X-1)-time. Similarly, the first likelihood P(2, X) and the first
likelihood P(3, X) are obtained based on the organized code Ja(2),
the recursive organized code Sa(2) and the second likelihood Ra(2,
X-1) at the (X-1)-time and the organized code Ja(3), the recursive
organized code Sa(3) and the second likelihood Ra(3, X-1) at the
(X-1)-time, respectively at steps 110 and 120.
[0020] Next, the interleaver 65 changes the order of the first
likelihood P(1, X), P(2, X) and P(3, X) at step 130. Hereinafter,
the first likelihood P(1, X), P(2, X) and P(3, X) are referred to
as the first likelihood Pa(1, X), Pa(2, X) and Pa(3, X),
respectively.
[0021] Next, the soft-determination decoder 63 obtains the second
likelihood R(1, X) based on the organized code Jb(1), the recursive
organized code Ua(1) and the first likelihood Pa(1, X-1).
Similarly, the second likelihood R(2, X) and the second likelihood
R(3, X) are obtained based on the organized code Jb(2), the
recursive organized code Ua(2) and the first likelihood Pa (2, X-1)
and the organized code Jb(3), the recursive organized code Ua(3)
and the first likelihood Pa(3, X-1), respectively at steps 140 and
150.
[0022] Next, the deinterleaver 66 changes the order of the second
likelihood R(1, X), R(2, X) and R(3, X) so as to correspond to the
order of the organized code J(1), J(2) and J(3) at step 160.
Hereinafter, the order-changed second likelihood R(1, X), R(2, X)
and R(3, X) are referred to as second likelihood Ra(1, X), Ra(2, X)
and Ra(3, X), respectively.
[0023] As described above, the accuracy of the second likelihood
Ra(i, X) is improved by repeating the calculation for the second
likelihood Ra for X times. The soft-determination portion 67
obtains the decoded data from the second likelihood Ra(i, X).
Therefore, the decoded data becomes extremely precise. Accordingly,
it is precisely determined which "0" or "1" is sent as the
information symbol.
[0024] The repeating time X is fixed to a predetermined value in
consideration of worse condition in communication environment
between the radio wave transmitter and the radio wave receiver.
However, the communication environment varies due to various
reasons. Therefore, this fixed repeating of the calculation results
in excess calculation as waste of process when the communication
environment is good. This excess calculation lowers the process
speed and efficiency.
SUMMARY OF THE INVENTION
[0025] An object of the present invention is to provide a decoder
capable of decoding efficiently and a receiver using the same.
[0026] According to a first aspect of the invention, a decoder has
a decoding portion for decoding a received error-correction code
using a parameter, an error calculating portion for calculating an
error level of a known code, a parameter determining portion for
determining the parameter based on the calculated error level.
[0027] The error level of the known code can denote the
communication environment, and is a basis for the parameter.
Therefore, the parameter is associated to the communication
environment. Therefore, the received code can be decoded
effectively by using this parameter.
[0028] Preferably, it is determined whether or not the error level
is high as compared to a reference level. When the error level is
high, the first value is used as the parameter. When the error
level is not high, the second value is used as the parameter so
that the decoding accuracy becomes low but the decoding speed
becomes high.
[0029] The decoding portion may calculate the likelihood of the
information code for a predetermined number of times based on the
error-correction code and the likelihood of the previous-time
information code, and obtain the decoded data based on the
calculated likelihood of the information code.
[0030] The determining portion may regard the predetermined number
of times as the parameter.
[0031] Namely, a repeating number of times for the calculation of
the likelihood can be set to a value suitable for the communication
environment by determining the repeating times based on the error
level of the known code.
[0032] According to a second aspect of the present invention, a
decoder, which is adopted to a receiver which digital-converts an
error-correction code to an N-bit digital signal, has an error
calculating portion for calculating an error level of a know code,
a determining portion for determining whether or not the error
level of the know code is equal to or higher than a reference
value, a data setting portion for obtaining the digital data which
has modified content of the N-bit digital signal so that L (L<N)
bits of the N-bit digital signal from a LSB side of the signal are
fixed to predetermined values, and a decoding portion performing a
decode process based on the N-bit digital signal when the error
level is equal to or higher than the reference level and performing
a decode process based on the data obtained by the data setting
portion when the error level is lower than the reference level.
[0033] With this feature, the signal is decoded in association with
the error level so that the process speed of decoding can be faster
when the level is lower than the reference level.
[0034] The data setting portion can be a data converting portion
which converts the N-bit digital signal to a M-bit (M<N) digital
data so that the process speed of decoding can be faster by
performing the decode using the M-bit digital signal when the error
level is lower than the reference level.
[0035] The above-described decoders can be adopted to a turbo
decoder and to a receiver.
[0036] Other features and advantages of the resent invention will
become more apparent from the following detailed description made
with reference to the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0037] FIG. 1 is a block diagram showing a turbo encoder of the
related art;
[0038] FIG. 2 is a block diagram showing a turbo decoder of the
related art;
[0039] FIG. 3 is a flow chart of the turbo decoder in FIG. 2;
[0040] FIG. 4 is a structure of a radio wave communication system
of a first embodiment of the present invention;
[0041] FIG. 5 is a flow chart of a control portion in FIG. 4;
[0042] FIG. 6 is a flow chart of a control portion of a second
embodiment of the present invention;
[0043] FIG. 7A is a sample of an N-bit digital data in the second
embodiment of the present invention;
[0044] FIG. 7B is a sample of a modified N-bit digital data second
embodiment of the present invention;
[0045] FIG. 7C is a sample of a M-bit digital data second
embodiment of the present invention;
[0046] FIG. 8 is a block diagram showing a turbo decoder of a third
embodiment of the present invention; and
[0047] FIG. 9 is a flow chart of a control portion in FIG. 8.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
[0048] Specific embodiments of the present invention will now be
described hereinafter with reference to the accompanying drawings
in which the same or similar component parts are designated by the
same or similar reference numerals.
[0049] As shown in FIG. 4, a radio wave communication system has a
transmitter 200 and a receiver 300. The transmitter 200 has a turbo
encoder 210, a pilot adder 220, a modulator 230 and an antenna 240.
The turbo encoder 210 has the same structure of the turbo encoder
as shown in FIG. 1, that is, it produces an organized code J(i),
recursive organized codes S(i) and U(i), and converts these codes
to an error-correction code as the serial data.
[0050] The pilot adder 220 adds a pilot symbol (P: a known symbol
as a known code) to the error-correction code of the serial data,
and outputs it. The modulator 230 modulates the output signal from
the pilot adder 220 and transmits it from the antenna 240.
[0051] The receiver 300 has an antenna 310, an A/D converter 320, a
demodulator 330 and a turbo decoder 340. The antenna 310 and a
receiver circuit (not shown) receive a transmitted signal from the
transmitter 200. The A/D converter 320 converts the received signal
as an analog signal into a digital signal.
[0052] The demodulator 330 demodulates the digital data output from
the A/D converter 320 so that a N-bit pilot symbol and a N-bit
error-correction code are obtained. Thus, an N-bit organized code
Ja(i), an N-bit recursive organized code Sa(i), an N-bit recursive
organized code Ua(i) are obtains.
[0053] The turbo decoder 340 has a BER measuring portion 341, a
pilot removing portion 342, a control portion 343 and a decoding
portion 344. The BER measuring portion 341 obtains a BER (Bit Error
Rate) of the pilot symbol output from the A/D converter 320.
[0054] The pilot removing portion 342 removes each pilot symbol
from output signals sent from the A/D converter 320, and then,
outputs an error-correction code of a block data. The control
portion 343 includes a microcomputer to determine a repeating
number of times for a calculation of a second likelihood Ra, and
orders the decoding portion 344 to decode using this repeating
number of times. The decoding portion 344 has the same structure of
the turbo decoder 60 as shown in FIG. 1, that is, it obtains the
second likelihood Ra based on the organized code J(i), the
recursive organized codes S(i) and U(i), and obtains a decoded data
based on this second likelihood Ra.
[0055] Hereinafter, a process for determining the repeating number
of times of the calculation for the second likelihood Ra will be
described with reference to a flow chart shown in FIG. 5. The
control portion 343 performs a computer program in association with
the flow chart shown in FIG. 5.
[0056] First of all, it is determined whether or not the BER (error
rate) of the pilot symbol output from the BER measuring portion 341
is equal to or higher than a reference value (for example, 50%) at
step 400. When the BER is equal to or higher than the reference
value, the repeating number of times is set at A times at step 410.
In this embodiment, the A times is obtained previously by
performing a simulation in consideration of a bad condition of the
communication environment.
[0057] Next, the calculation for the second likelihood Ra(i, X) is
repeated for A times to obtain the second likelihood Ra(i, A), and
the decoded data is calculated based on the second likelihood Ra(i,
A) by the decoding portion 344 at step 420.
[0058] On the other hand, when the BER output from the BER
measuring portion 341 is determined to be lower than the reference
value at step 400, the repeating number of times is set at B times
that is smaller than A at step 430. In this embodiment, the A times
is obtained previously by performing a simulation in consideration
of a good condition of the communication environment.
[0059] Next, the calculation for the second likelihood Ra(i, X) is
repeated for A times to obtain the second likelihood Ra(i, B), and
the decoded data is calculated based on the second likelihood Ra(i,
B) by the decoding portion 344 at step 420.
[0060] As described above, the BER measuring portion 341 obtains
the BER of the pilot symbol. Then, the control portion 343
determines the repeating number of times of the calculation for the
second likelihood Ra(i, X) as a parameter. After that, the decoding
portion 344 decodes the signal with the number of times of the
calculation.
[0061] As described above, the repeating number of times of the
calculation for the second likelihood Ra(i, X) can be obtained in
association with the condition of the communication environment.
Therefore, the error-correction code can be decoded efficiently by
using this various repeating number of times. Moreover, the
repeating number of times of the calculation for the second
likelihood Ra(i, X) is varied according to the BER of the pilot
symbol so that the error-correction code can be decoded while the
precision in calculation for the second likelihood Ra(i, X), or the
precision in decoding is restrained from deteriorating.
[0062] In the first embodiment, the decoder is the turbo decoder.
However, this technique can be adopted to a decoder which performs
the repeating calculation.
[0063] (Second Embodiment)
[0064] In the first embodiment, the example in which the
error-correction code output from the pilot removing portion 342 is
input to the decoding portion 344 to be decoded without changing is
described. However, the error-correction code output from the pilot
removing portion 342 can be changed.
[0065] A data modification process in the control portion 343 will
be described with reference to a flow chart shown in FIG. 6. The
data modification process will start when the N-bit
error-correction code (digital data) is input from the pilot
removing portion 342 to the control portion 343.
[0066] First, when it is determined that the BER is equal to or
higher than the reference value, the decoder recognizes the
communication environment is in bad condition, and performs the
following process.
[0067] For example, when the N-bit error-correction code having a
structure as shown in FIG. 7A as a content of the block data is
input to the control portion 343, the control portion 343 inputs
the N-bit error-correction code to the decoding portion 344 without
changing it at step 410A. The decoder decodes this original
error-correction code at step 420.
[0068] On the other hand, when it is determined that the BER is
lower than the reference value, the decoder recognizes the
communication environment is in bad condition, and performs the
following process.
[0069] For example, when the N-bit error-correction code having a
structure as shown in FIG. 7A as a content of the block data is
input to the control portion 343, the control portion 343 sets each
data in the L-bit data (L<N) of the N-bit error-correction code
from its LSB side to a fixed constant value (for example, "0") to
produce the N-bit error-correction code as shown in FIG. 7B. The
production of the new N-bit error-correction data which has fixed
value in the content of its L-bit data from the LSB side is
performed every N-bit error-correction data in a block data. The
error-correction codes are decoded by the decoding portion 344 at
step 420 based on all of the N-bit error-correction codes for the
block data set at step 430B.
[0070] As described above, when the BER from the BER measuring
portion 431 is lower than the reference value to thereby
determining the communication environment is in good condition, the
N-bit error-correction code whose L-bit data is fixed to the
constant value is used to be decoded at the decoding portion 344.
Therefore, the process speed becomes higher than that of the case
using the N-bit error-correction code having data just sent from
the pilot removing portion 342, i.e., the original N-bit
error-correction code after being removed its pilot symbol at the
pilot removing portion 342.
[0071] Although the error-correction code that have the constant
value in the L-bit data from its LSB side has low precision as
compared to the original error-correction code which is not
changed, the error-correction code for decoding is changed in
association with the BER of the pilot symbol so that the decoding
is performed effectively while the precision of the decoding is
restrained from deteriorating.
[0072] (Third Embodiment)
[0073] In the second embodiment, the example in which the N-bit
error-correction code that have the constant value in the L-bit
data from its LSB side is used for decoding is described. In this
embodiment, a M-bit error-correction code whose size is smaller
than the N-bit error-correction code (i.e., M<N) is produced
from the N-bit error-correction code output from the pilot removing
portion 342, and this M-bit error-correction code is used for
decoding.
[0074] A turbo decoder 340 for this case is shown in FIG. 8. The
turbo decoder 340 has a BER measuring portion 341, a pilot removing
portion 342, a control portion 343A, and decoding portions 344A and
344B. Incidentally, the numerals same as those shown in FIG. 4
denotes the same or similar portions in FIG. 4.
[0075] The pilot removing portion 342 removes plural pilot symbols
of the signals output from the A/D converter 320 to obtain only
error-correction codes. In this embodiment, an error-correction
output from the pilot removing portion 342 is an N-bit signal, and
also digital data having an E(integer) number of decimals.
[0076] The decoding portion 344A obtains the decoded data based on
an N-bit error-correction code having the E number of decimals
while the decoding portion 344B obtains the decoded data based on a
M-bit error-correction code having a F(integer) number of decimals.
Wherein, the bit-number M is smaller than the bit-number N
(M<N), the integer F is smaller than the integer E (F<E).
[0077] The control portion 343A receives the N-bit error-correction
code from the pilot removing portion 342, and lets the decoding
portions 344A and 344B decode the N-bit error-correction code or
the M-bit error-correction code in association with the N-bit
error-correction code. A detail flow of this process will be
described with reference to FIG. 9.
[0078] First, when it is determined that the BER is equal to or
higher than the reference value at step 400, the N-bit
error-correction code for the block data that is not changed and
sent from the pilot removing portion 342 is decoded by the decoding
portion 344A at steps 410A and 420.
[0079] On the other hand, when it is determined that the BER is
lower than the reference value, the control portion 343A performs
the following process.
[0080] For example, when the N-bit error-correction code for the
block data as shown in FIG. 7A is input to the control portion 343A
from the pilot removing portion 342, the control portion 343A
modifies the N-bit error-correction code to the M-bit
error-correction code as shown in FIG. 7C. This modification means
that the error-correction code having the E number of decimals is
modified to the error-correction code having the E number of
decimals at step 430C. After that, the modified error-correction
code for the block data is decoded by the decoding portion 344A at
step 420.
[0081] As described above, when the BER from the BER measuring
portion 431 is lower than the reference value to thereby
determining the communication environment is in good condition, the
error-correction code having the F number of decimals is used for
decoding at the decoding portion 344. Although the error-correction
code having the F number of decimals has low precision as compared
to the original error-correction code having the E number of
decimals, the error-correction code for decoding is changed in
association with the BER of the pilot symbol so that the decoding
is performed effectively while the precision of the decoding is
restrained from deteriorating.
[0082] Incidentally, the turbo decoder is employed in the second
and third embodiment, the other type of the decoder can be
used.
[0083] While the present invention has been shown and described
with reference to the foregoing preferred embodiment, it will be
apparent to those skilled in the art that changes in form and
detail may be therein without departing from the scope of the
invention as defined in the appended claims.
* * * * *